Heimdall++: Optimizing GPU Utilization and Pipeline Parallelism for Efficient Single-Pulse Detection

2025-11-29

AI Summary

Modern radio telescopes demand high-throughput, low-latency real-time single-pulse detection under high time–frequency resolution. However, the widely adopted GPU-accelerated tool Heimdall suffers from low GPU utilization and pipeline stalls due to its sequential execution model and resource contention in intermediate stages. To address this, we propose an efficient parallelization framework featuring fine-grained GPU kernel concurrency, CPU–GPU task decoupling, multi-threaded pipelined scheduling, and optimized memory access—collectively mitigating computational idle time. The framework maintains full output-format compatibility with Heimdall. Evaluations on an NVIDIA RTX 3080 Ti demonstrate a 2.66× speedup for single-file processing and a 2.05× speedup for batched multi-file processing, with zero deviation in detection results.

Technology Category

• Machine Learning: Hardware-aware ML
• Search and Optimization: Distributed Search
• Data Mining & Knowledge Management: Scalability, Parallel & Distributed Systems

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Data management and stream processing for Web, mobile and wireless applications
• Search and Retrieval-Augmented AI: Efficiency and scalability of Web search engines
• Security and Privacy: Large-scale security measurements

Abstract

With the increasing time and frequency resolution of modern radio telescopes and the exponential growth in observational data volumes, real-time single-pulse detection has become a critical requirement for time-domain radio astronomy. Heimdall, as a representative GPU-accelerated single-pulse search tool, offers substantial performance advantages over CPU-based approaches. However, its sequential execution model and resource contention in intermediate processing stages limit GPU utilization, leading to suboptimal throughput and increased computational latency. To address these limitations, we present Heimdall++, an optimized successor to Heimdall that incorporates fine-grained GPU parallelization, enhanced memory management, and a multi-threaded framework to decouple CPU-bound and GPU-bound processing stages. This design mitigates the GPU stall problem and improves end-to-end efficiency. We evaluated Heimdall++ on a system equipped with NVIDIA RTX 3080 Ti GPUs using both a single large-scale observational file and multiple files. Experimental results demonstrate that Heimdall++ achieves up to 2.66x speedup in single-file processing and 2.05x speedup in multi-file batch processing, while maintaining full consistency with the original Heimdall's search results.

Problem

• Optimizing GPU utilization for single-pulse detection
• Reducing computational latency in radio astronomy pipelines
• Improving throughput via fine-grained GPU parallelization

Innovation

• Fine-grained GPU parallelization enhances processing efficiency
• Multi-threaded framework decouples CPU and GPU processing stages
• Enhanced memory management mitigates GPU stall and improves throughput